RTD vs Thermocouple: How to Choose
8 min read · updated 2026-07-12
The two workhorses of industrial temperature measurement solve the same problem with opposite physics — and opposite strengths. Here's the practical decision guide.
How each one works
An RTD (resistance temperature detector) exploits the fact that platinum's electrical resistance rises predictably with temperature — 100 Ω at 0 °C for a PT100, 138.51 Ω at 100 °C, following the Callendar–Van Dusen equation standardised in IEC 60751. You measure a resistance and compute a temperature — precisely what the PT100/PT1000 calculator does, both directions, with tolerance classes.
A thermocouple exploits the Seebeck effect: a junction of two dissimilar metals generates a tiny voltage that depends on temperature. It measures the difference between its hot junction and the reference (cold) junction — which is why every reading needs cold junction compensation, handled properly in the thermocouple mV ↔ temperature calculator for all eight standard types.
The comparison that matters
| RTD (PT100) | Thermocouple | |
|---|---|---|
| Accuracy | ±0.15–0.8 °C typical (Class A/B) | ±1–2.5 °C typical (Class 1/2) |
| Range | −200 to +850 °C (practically ~600) | −270 to +1,820 °C by type |
| Response speed | Slower (more thermal mass) | Faster, especially exposed junctions |
| Long-term drift | Low, very stable | Higher, degrades with thermal cycling |
| Ruggedness | More fragile (fine element) | Very rugged, vibration-tolerant |
| Wiring | Copper OK, but 2/3/4-wire matters | Needs matching extension cable |
| Relative cost | Sensor higher, wiring cheap | Sensor cheap, special cable adds up |
The wiring gotchas — where installations go wrong
RTDs: the measurement is a resistance, so every ohm of lead wire adds error — about 2.6 °C per ohm on a PT100 in a 2-wire circuit. Three-wire connection compensates matched leads (industry default); four-wire eliminates lead resistance entirely (lab-grade); or use a PT1000, which shrinks the same error ten-fold.
Thermocouples: the signal is microvolts, and any junction of dissimilar metals creates new thermocouples. Ordinary copper wire between the sensor and the panel silently moves the reference junction to wherever the copper starts — at an unknown temperature. Matching extension or compensating cable must run all the way to the point where cold junction compensation happens. Reversed polarity is the other classic: roughly right at ambient, increasingly wrong as the process heats.
Decision guide
Choose an RTD when: accuracy and stability lead the requirements — custody metering, batch reactors, pharma, energy balance points, anything below ~500 °C where ±0.5 °C matters and drift is unacceptable.
Choose a thermocouple when: the temperature exceeds RTD territory (furnaces, exhausts, kilns — types K/N to ~1,200 °C, R/S/B beyond), when the environment is high-vibration, when response speed matters more than half a degree, or when sensors are consumables replaced often and cost rules.
A useful plant-level heuristic: RTDs where you control precisely, thermocouples where you survive harshly. Both ultimately feed a transmitter producing 4–20 mA — ranged with the same scaling math as any loop (see the 4–20 mA calculator), and converted between °C/°F/K with the temperature unit converter.
The bottom line
Below 500 °C and accuracy-driven: RTD, wired 3- or 4-wire. Above, or rugged, or fast, or cheap-per-point: thermocouple, with the right extension cable and honest expectations about drift. The worst choice is the unexamined default — both sensors fail quietly when their wiring rules are ignored.
Calculators used in this guide: PT100/PT1000 ↔ temperature · thermocouple mV ↔ temperature · °C ↔ °F ↔ K ↔ °R